Explore adaptation enhancements to feature_extraction.py

[kharaldsson]

• [x] Word2Vec / Embeddings
• [ ] Sentiment Analysis

[kharaldsson]

Ok. No changes necessary to Word2Vec approach. We don't use pre-trained embeddings for english--we train a Word2Vec model on the fly and use those embeddings. Thus, since we're training a model on the fly, we can keep the same approach we're using now.

[kharaldsson]

Sentiment analysis is another matter. Some options:

• Switch from vader to vader-multi (https://pypi.org/project/vader-multi/)
• Try a fit-for-purpose analyzer Spanish: https://pypi.org/project/sentiment-analysis-spanish/
• Translate using TextBlob first.